Structure–Activity Relationships for Carcinogenic Potential

Two important disadvantages of long-term animal bioassays are that testing involves substantial amounts of time and money, and that high doses are usually used in the testing process. These disadvantages can be circumvented using (quantitative) structure-activity relationships ((Q)SARs). In the field of computational toxicology, (Q)SARs are predictive models that provide a quantitative measure of the relationship between the chemical structure and a measure of a given health-related end point. Such relationships can be expressed in terms of continuous dose-response data (e.g. carcinogenic potency) based on some type of regression analysis for quantitative end points, or a dichotomous classification (e.g. yes/no-type answers for carcinogenicity, etc.) based on discriminant analysis or other pattern recognition techniques for qualitative end points. There are a limited number of (Q)SAR models to predict the carcinogenicity of various chemicals, a majority of which relate the carcinogenic potency to measures of carcinogenicity such as mutagenicity, lethal dose (LD50) or the maximum tolerated dose (MTD). Other (Q)SAR models relate the carcinogenicity of a chemical to its structure, either in terms of its chemical fragments (groups of one or more atoms that make up the chemical structure) or in terms of its physical and chemical properties. In addition, a variety of commercial and noncommercial software that contain modules to predict the carcinogenicity of chemicals are also available. Keywords: structure-activity relationship; quantitative structure-activity relationship; SAR; QSAR; carcinogenicity; cancer; carcinogenic potency; toxicity prediction; computer modelling; tumour; tumourigenicity; oral slope factor; tumourigenicity dose rate